Transcriptional regulation of osteopontin gene in vivo by ... - Nature

Oncogene (1998) 17, 1517 ± 1525  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc

Transcriptional regulation of osteopontin gene in vivo by PEBP2aA/CBFA1 and ETS1 in the skeletal tissues Motohiko Sato1,2, Eiichi Morii1, Toshihisa Komori3, Hirohisa Kawahata1, Mizuo Sugimoto1, Kunihiro Terai1, Hideo Shimizu1, Takahiro Yasui1,4, Hideki Ogihara1, Natsuo Yasui2, Takahiro Ochi2, Yukihiko Kitamura1, Yoshiaki Ito5 and Shintaro Nomura1 1

Department of Pathology, Osaka University Medical School, Suita, Osaka 565-0871; 2Department of Orthopaedic Surgery, Osaka University Medical School, Osaka 565 ± 0871; 3Department of Medicine III, Osaka University Medical School, Suita, Osaka 565 ± 0871; 4Department of Urology, Nagoya City University Medical School, Nagoya, Aichi 467-8601; 5Department of Viral Oncology, Institute for Virus Research, Kyoto University Medical School, Kyoto, 606 ± 8397, Japan

Osteopontin (Opn) and polyoma enhancer-binding protein (PEBP) 2aA/core binding factor (CBFA) 1 have been suggested to play important roles in ossi®cation. The overlapping localization of opn and PEBP2aA/ CBFA1 mRNA, and the marked decrease of opn mRNA expression in PEBP2aA knockout mice, indicated that the transcription of opn gene was controlled by PEBP2aA. In the present study, we determined the direct regulation of PEBP2aA on the opn promoter activity. Opn promoter activity was markedly enhanced by PEBP2aA and ETS1 in a synergistic manner. The synergistic eect was diminished when either the PEBP2aA or ETS1 binding site was mutated, or the spatial arrangement of these sites was mutated by a 4-nt insertion. The distance between these sites was important for transactivation but not protein-DNA binding. The direct interaction between PEBP2aA and ETS1 was depended on protein-DNA binding. These results suggested that the speci®c spatial arrangement of both sites and direct interaction between PEBP2aA and ETS1, were essential for promoter function. Furthermore, endogenous opn mRNA was decreased with the introduction of dominant negative PEBP2aA to MC3T3/ E1 cells expressing endogenous PEBP2aA, ETS1 and opn. These ®ndings suggest that PEBP2aA and ETS1 cooperate in vivo to regulate expression of the opn gene in the skeletal tissue. Cell type-speci®c regulation of Opn gene expression will also be discussed. Keywords: osteopontin; PEBP2aA/CBFA1; synergistic transactivation; osteogenesis

ETS1;

Introduction Osteopontin (Opn) is a major non-collagenous bone matrix protein. It is a sialic acid-rich phosphorylated glycoprotein originally isolated from the bone. The opn gene contains a Gly-Arg-Gly-Asp-Ser (GRGDS) -encoding motif that promotes cell attachment via avb3 integrin (Oldberg et al., 1986) and CD44 (Weber et al., 1996). Opn has been suggested to play important roles in the process of neoplastic metastasis (Brown et al., 1994; Oates et al., 1996) by enhancement of cell

Correspondence: S Nomura Received 20 February 1998; revised 20 April 1998; accepted 21 April 1998

attachment through Opn receptors. In addition to the integrin binding consensus sequence, Opn contains several additional putative functional domains; a stretch of nine or ten aspartic acid residues that could play a role in hydroxyapatite binding, two heparin binding domains, and a calcium binding domain (Denhardt and Guo, 1993). Previously, we investigated the expression of opn mRNA in the process of mouse development (Nakase et al., 1994; Nomura et al., 1988). In the skeletal tissue, opn mRNA is expressed in a population of osteoblasts, hypertrophic chondrocytes and osteoclasts in the late stages of embryogenesis. In addition, opn mRNA is expressed in a population of macrophages (Singh et al., 1991), T lymphocytes (Patarca et al., 1989), epithelial cells of distal kidney tubules (Nomura et al., 1988) and sensory hair cells in the inner ear (Takemura et al., 1994). The level of opn mRNA expression in the osteoblasts decreases after birth, and no detectable expression is observed in the osteogenic cells of mature bone. In contrast, opn expression is markedly enhanced in the callus formation process during fracture healing (Hirakawa et al., 1994) and distraction osteogenesis (Sato et al., in press). The possible involvement of Opn in calci®cation is suggested not only in physiological but also in pathological calci®cation. We demonstrated that opn mRNA-expressing cells are located adjacent to calci®ed areas in atherosclerosis (Hirota et al., 1993), breast cancer (Hirota et al., 1995) and renal stones (Kohri et al., 1993). Furthermore, these calci®ed areas were strongly reactive to Opn antibody. Although the close relationship between the presence of Opn and biological calci®cation has been extensively studied in vivo, the factors which regulate expression of the opn gene in vivo are still unclear. Opn mRNA has been detected in speci®c cell types and at speci®c times, so appropriate tissue-speci®c and timing-speci®c transcription factors might regulate expression of the opn gene. In addition, several transcription factors may synergistically regulate the tissue- and developmental stagespeci®c opn gene expression. Insight into the mechanism of tissue- or developmental stage-speci®c transcriptional activation requires identi®cation of the protein components that make up complex and the de®nition of both the protein-protein and protein-DNA interactions. The promoter region of opn contains several sites to which transcription factors are expected to bind (Denhardt and Guo, 1993; Guo et al., 1995). The major candidates for the transcription factors which

Transcriptional regulation of osteopontin gene in vivo M Sato et al

1518

might aect expression of the opn gene are steroid hormone receptors including vitamin D receptor, AP-1 (fos/jun), bHLHs, ets family and runt family etc. The promoter region up to 7204 nt is conserved among mouse, porcine and human homologues. In this region, binding sites of the runt family and ets family are located with 9 nt interval. Among the members of the runt and ets family, mRNA encoding polyoma enhancer-binding protein (PEBP) 2aA; also called core binding factor (CBF)A 1 (Komori et al., 1997) and ETS1 was detected in osteogenetic cells (Kola et al., 1993). We investigated transcriptional regulation of the opn gene by PEBP2aA/CBFA1 and ETS1 in vivo and in vitro. PEBP2aA/CBFA1 is a transcription factor that belongs to the runt family. Three runt domain genes, PEBP2aA/CBFA1, PEBP2aB/CBFA2 and PEBP2aC/ CBFA3, have been reported (Bae et al., 1993, 1995; Levanon et al., 1994; Ogawa et al., 1993a; Wijmenga et al., 1995). These molecules share a conserved 128amino acid domain called the runt domain because it was initially discovered in the Drosophila pair rule gene runt (Kagoshima et al., 1993). This domain mediates DNA binding as well as dimerization to PEBP2b/CBFB, which has no direct DNA binding ability, and confers enhanced DNA binding capacity to PEBP2a (Ogawa et al., 1993b; Wang et al., 1993). PEBP2a speci®cally recognizes a consensus sequence, PuACCPuCA and has been reported to control polyoma virus enhancer (Kamachi et al., 1990), murine leukemia virus enhancers (Wang and Speck, 1992), T cell-speci®c genes (Giese et al., 1995; Hallberg et al., 1992; Hsiang et al., 1993; Prosser et al., 1992; Redondo et al., 1992), enzymes (myeloperoxidase, neutrophil elastase, granzyme B serine protease) (Nuchprayoon et al., 1994; Wargnier et al., 1995) and cytokines and their receptors (GM-CSF, IL3, CSF-1) (Cameron et al., 1994; Frank et al., 1995; Takahashi et al., 1995; Zhang et al., 1994). PEBP2aA and ETS1 have been shown to bind together and interact synergistically on the promoter of the T cell receptor a gene (Giese et al., 1995) and T cell receptor b gene (Wotton et al., 1994). Recently, we generated PEBP2aA-de®cient mice by an insertional mutation (Komori et al., 1997). Homozygous (7/7) mice could not breath and died soon after birth due to defect of calci®cation in the ribs. The homozygotes were severely impaired in both endochondral and membranous ossi®cation. Histological and molecular examination of homozygotes demonstrated that the dierentiation of osteoblastic cells was almost completely blocked, whereas chondrogenesis was almost normal. The expression of bone matrix proteins mRNA including osteonectin, matrix Gla-protein, osteocalcin and opn has been investigated (Komori et al., 1997). Marked decreases in opn expression were found in hypertrophic chondrocytes and osteoblasts. These results suggested that the expression of opn mRNA is regulated by PEBP2aA in vivo, but it is still not clear whether PEBP2aA regulates opn gene expression directly or indirectly. In the present study, to investigate whether PEBP2aA directly regulates opn gene expression, we determined the eects of PEBP2aA on transactivation of the opn promoter, speci®c binding of PEBP2aA to the opn promoter region, and co-localization of PEBP2aA and

opn mRNA. We also examined the eects of dominant negative PEBP2aA on the expression of the opn gene in osteogenic cells. Furthermore, we also studied the synergistic eect of ETS1, mRNA of which was detected in osteogenic cells, and PEBP2aA on transactivation of the opn promoter. We demonstrated that PEBP2aA and opn mRNA have the same localization and PEBP2aA is essential for expression of the opn gene and ossi®cation. In addition, direct protein-protein contact between PEBP2aA and ETS1 is a prerequisite for the formation of a functional complexes and this interaction is facilitated by the

a

PORCINE

PORCINE

PORCINE

PORCINE

PORCINE

b

Figure 1 (a) Nucleotide sequences of the promoter regions of mouse, porcine and human osteopontin genes. The nucleotide numbers correspond to the mouse sequence. Asterisks have been inserted in sequences to maximize alignment. Bars in porcine and human sequences indicate nucleotides conserved in the mouse sequence. Runt and ets binding sites identi®ed in this study are boxed. (b) Nucleotide sequences of wild-type (oligo 1, 2, 3) and mutant (oligo 3 mut R, 3 mut E, 3 +1, 3 +4) forms of synthetic oligonucleotides within the mouse opn promoter. Oligo 1, 2, 3 spanned runt, ets, both runt and ets binding sites, respectively. Oligo 3 mut R had mutations in the runt binding site. Oligo 3 mut E had mutations in the ets binding site. Oligo 3 +1 and +4 had inserts of 1 and 4 nucleotides, respectively, between runt and ets binding sites


speci®c arrangement of the transcription factor binding sites within the opn promoter.

Results Conservation of nucleotide sequences in the promoter region of the opn gene Several putative binding sites such as those for Vitamin D receptor, AP1, runt family, ets family and bHLHs etc. are located in the promoter region of opn (Craig and Dehardt, 1991; Guo et al., 1995). Particularly, the nucleotide sequence up to position 7204 is highly

conserved among mouse, porcine and human homologues, where the runt family and ets family binding sites are located (Figure 1a). The runt and ets sites are separated by a stretch of 9 nt which are conserved between these species. Overlapped expression of PEBP2aA and Opn gene Opn and PEBP2aA mRNA-expressing cell types were examined in wild-type mice. Paran sections containing long bones of +/+ mice embryos 18.5 day post coitum (p.c.) were hybridized with opn and PEBP2aA cRNA probes. Strong opn mRNA signals were observed in the osteoblasts and hypertrophic chon-

Figure 2 The histology of d18.5 tibia in normal (+/+) embryos. (a)*(f) are sequential sections. (a) Hematoxylin and eosin staining. (b) Von Kossa's staining. Calci®ed cartilage were seen in the diaphysis. (c) The section was hybridized with opn cDNA probe. Opn mRNA signal was seen in some hypertrophic chondrocytes at the zone of provisional calci®cation ($) and some osteoblasts in the newly formed bone. hc, hypertrophic chondrocyte; ob, osteoblast. (d) The section was hybridized with PEBP2aA cDNA probe. PEBP2aA mRNA signal was seen in some of hypertrophic chondrocytes in the zone of provisional calci®cation ($), some of osteoblasts in the newly formed bone, and premature chondrocytes. hc, hypertrophic chondrocyte; ob, osteoblast; pc, premature chondrocyte. (e) The section was hybridized with PEBP2aB cDNA probe. (f) The section was hybridized with PEBP2aC cDNA probe. Bar=40 mm

1519


1520

drocytes in limbs (Figure 2c). The adjacent sections were stained with von Kossa's stain, and distinct calci®cation signals were detected in the same regions as opn mRNA-expressing cells (Figure 2b). PEPB2aA mRNA was localized in the ®broblasts-lke cells around periosteal layer in the limbs, osteoblasts and chondrocytes (Figure 2d). Although the zones of expression of opn and PEBP2aA mRNA overlapped, PEBP2aA mRNA-expressing cells were more widely distributed than those expressing opn mRNA and the population of opn mRNA-expressing cells was included in that expressing PEBP2aA mRNA. Furthermore, localizations of PEBP2aB and PEBP2aC mRNA were also examined. No significant signals of these mRNAs were detected in the skeletal tissues (Figure 2e and f). Transactivation of the Opn promoter by PEBP2aA and PEBP2b We examined whether the AACCACA motif (from 7136 to 7130) in the upstream region of the opn gene mediated the transactivation eect of PEBP2aA. The promoter region and the ®rst exon of the opn gene (7910 to +66) was ligated upstream of the luciferase gene. The expression plasmid containing PEBP2aA cDNA was co-transfected into NIH3T3 ®broblasts with the luciferase construct containing the opn promoter. The co-expression of PEBP2aA increased the luciferase activity by approximately tenfold compared with control experiment (Figure 3a). Then, we produced reporter plasmids with deletions in the opn promoter. Deletion of the sequence between nt 7910 and 7255 (7254 luc) showed no eect on transactivation. However, deletion of the sequence between nt 7910 and 795 (794 luc) abolished the transactivation ability of PEBP2aA, suggesting that the site between nt 7254 and 795 which contained the AACCACA motif from nt 7136 to 7130 was essential for transactivation. To con®rm that this motif is the response site of PEBP2aA, 7254 luc construct in which the AACCACA motif was changed to AAGAACA (7254 runt-mut luc) was used as a reporter; 7254 runt-mut luc did not respond to PEBP2aA. These results indicate that the AACCACA motif of 7136 to 7130 in the upstream region of the opn gene acted as a response site to PEBP2aA. Furthermore, the eects of PEBP2b on the opn promoter activity were examined (Figure 3b). Promoter activity was increased by sevenfold by PEBP2b and transactivation was dependent on the presence of the AACCACA motif in reporter plasmid. When both PEBP2aA and PEBP2b cDNA were co-transfected, the promoter activity was markedly increased. Synergistic eect of PEBP2aA, PEBP2b and ETS1 on promoter activity of the opn gene The ets family (ETS1, PU.1/Spi-1) binding motif is located 9 bp downstream from the PEBP2aA binding motif in the promoter region of the opn gene. The transactivation eect of PEBP2aA and ETS1 on the opn promoter was examined. As reporter plasmids, 7254 luc, 7254 runt-mut luc and 7254 ETS-mut luc in which the GAGGAA motif (from 7120 to 7115) of 7254 luc was changed to GACCAA were used. Co-

transfection of 7254 luc and ETS1 increased the luciferase activity by eightfold (Figure 3c). In contrast, co-transfection of 7254 ETS-mut luc and ETS1 did not increase luciferase activity. Surprisingly, cotransfection of 7254 runt-mut luc and ETS1 also did not increase luciferase activity. These results indicated that the GAGGAA motif from nt 7120 to 7115 acts as a response site to ETS1 and in addition, the eect of ETS1 on opn promoter activity was possibly mediated by endogenous expression of member of the runt family in NIH3T3 cells. Northern blot analysis of PEBP2aA expression in NIH3T3 cells is shown in Figure 4. A synergistic eect on opn promoter activity was observed by co-transfection of PEBP2aA, PEBP2b and ETS1. Co-transfection of 7254 luc, PEBP2aA, a

b

c

Figure 3 (a) The eects of transient expression of PEBP2aA cDNA on the luciferase activity under the control of the osteopontin promoter starting from nt 7910, or deleted promoters. The deleted plasmid that started at nt 7254 or 794 was examined. The luciferase construct with a mutation in the AACCACA motif was also examined. Bars indicate the s.e. of three assays. (b) The synergistic eect of PEBP2b on the luciferase activity under control of the osteopontin promoter starting from nt 7254. Bars indicate the s.e. of three assays. (c) Luciferase assay by co-transfection of expression vector constructs, PEBP2aA, PEBP2b and ETS1. The eects of PEBP2aA or ETS1 binding site mutation on osteopontin promoter were examined. The luciferase construct with a 4 nt insert between PEBP2aA and ETS1 sites was also examined. Bars indicate the s.e. of three assays


PEBP2b and ETS1 increased the luciferase activity by 75-fold. In contrast, co-transfection of 7254 runt-mut luc, PEBP2aA, PEBP2b and ETS1 or 7254 ETS-mut luc, PEBP2aA, PEBP2b and ETS1 did not increase luciferase activity. To investigate the interaction between PEBP2aA and ETS1, 7254 mut +4 luc, in which four nucleotides were inserted between these sites in 7254 luc to alter the helical phasing relationship, was used as a reporter plasmid. Cotransfection of 7254 mut +4 diminished the synergistic eect of PEBP2aA, PEBP2b and ETS1. These results indicated that PEBP2aA, PEBP2b and ETS1 synergistically transactivated the opn promoter, and this synergistic eect required PEBP2aA and the ETS1 binding sites with the appropriate helical phasing relationship.

band was slower than that obtained when only PEBP2aA (Figure 5a, lane 2) or only ETS1 was incubated with oligo 3 (Figure 5a, lane 5). To characterize the band detected in lane 8, runt site mutated oligonucleotide (oligo 3 mut R) or ets site mutated oligonucleotide (oligo 3 mut E) were used.

Demonstration of the interaction between PEBP2aA and ETS1 The interaction of PEBP2aA and ETS1 was analysed by electrophoretic mobility shift assay (EMSA). Oligonucleotides used in the present study are shown in Figure 1b. When PEBP2aA and ETS1 were incubated with an 32P-labeled oligonucleotide containing runt and ets binding sites (oligo 3), a strong signal was detected (Figure 5a, lane 8). The mobility of the

Figure 4 Northern blotting analysis of the expression of the PEBP2aA and ETS1 in NIH3T3 and MC3T3/E1 cells. RNA was extracted from these cells and 75 mg of total RNA was loaded per lane

b

a

Figure 5 (a) Cooperative binding of PEBP2aA and ETS1 depended to the presence of both PEBP2aA and ETS1 binding sites. Radiolabeled oligonucleotides (oligo 3, containing both PEBP2aA and ETS1 binding sites) or mutants were incubated with PEBP2aA, ETS1 or both proteins together. Oligo 3 was incubated with PEBP2aA (lane 2), ETS1 (lane 5) or both proteins (lane 8). Formation of PEBP2aA-ETS1 complex with radiolabeled wild-type oligo 3 is shown in lane 8. The PEBP2aA site mutant (oligo 3 mut R) was incubated with PEBP2aA (lane 3), ETS1 (lane 6) or both proteins (lane 9). The ETS1 site mutant (oligo 3 mut E) was incubated with PEBP2aA (lane 4), ETS1 (lane 7) or both proteins (lane 10). The spacing mutants, oligo 3+1 (1 nt insertion between both sites) or oligo 3+4 (4 nt insertion), were incubated with both proteins (lanes 11 and 12). Speci®c complexes are indicated. (b) Direct binding of PEBP2aA and ETS1 in a DNA-dependent manner. Both PEBP2aA and ETS1 proteins were incubated with radiolabeled oligonucleotides, oligo 1 (lane 2), oligo 2 (lane 3), oligo 3 (lane 4) or both oligo 1 and oligo 2 (lane 5). Speci®c complexes are indicated

1521


1522

The mobility of the band obtained when PEBP2aA and ETS1 were incubated with the oligo 3 mut R (Figure 5a, lane 9) was identical to that obtained when ETS1 was incubated with oligo 3 (Figure 5a, lane 5) or oligo 3 mut R (Figure 5a, lane 6). Similarly, the mobility of the band obtained when PEBP2aA and ETS1 were incubated with oligo 3 mut E (Figure 5a, lane 10) was identical to that obtained when PEBP2aA was incubated with oligo 3 (Figure 5a, lane 2) or oligo 3 mut E (Figure 5a, lane 4). Therefore the band shown in Figure 5a, lane 8 was a complex of PEBP2aA-ETS1-oligo 3. A band with the same mobility was obtained when PEBP2aA and ETS1 were incubated with oligo 3+1 (Figure 5a, lane 11) or oligo 3+4 (Figure 5a, lane 12), which was created by inserting one or four nt between runt and ets sites in oligo 3. These results demonstrated that PEBP2aA and ETS1 bound together to DNA, but did not demonstrate direct binding between PEBP2aA and ETS1. To demonstrate the direct binding of PEBP2aA-ETS1, we used two oligonucleotides which contained either the runt binding site (oligo 1) or ets binding site (oligo 2). When PEBP2aA and ETS1 were incubated with the oligo 1 and oligo 2, three distinct bands were obtained (Figure 5b, lane 5). The mobility of the faster band was the same as that obtained when PEBP2aA and ETS1 were incubated with oligo 2 (Figure 5b, lane 3). The mobility of the middle band was the same as that obtained when PEBP2aA and ETS1 were incubated with oligo 1 (Figure 5b, lane 2). The mobility of the slower band was the same as that obtained when PEBP2aA and ETS1 were incubated with oligo 3 (Figure 5b, lane 4). These results indicated the presence of oligo 1-PEBP2aA-ETS1oligo 2 complex and direct binding of PEBP2aA and ETS1. Dominant negative eect of runt domain (pCX2neoNH) on the expression of the endogenous Opn gene in the osteogenic cell line To investigate whether expression of the endogenous opn gene was regulated by PEBP2aA, the eects of transfected pCX2neoNH encoding a runt domain which had a dominant negative eect of PEBP2aA on the expression of opn mRNA were examined. When NIH3T3 cells were transfected with pCX2neoNH, opn promoter activity was markedly decreased (data not shown). Osteoblastic cell line MC3T3/E1 cells expressed PEBP2aA and ETS1 mRNA (Figure 4). We examined how the wild type or mutated luciferase reporters responded to endogenous PEBP2aA and ETS1 in MC3T3/E1 cells. Transfection of wild type luciferase reporter (7254 luc) increased the luciferase activity by ®vefold compared with the transfection of backbone vector pSPLuc harboring no opn sequence. However, mutated luciferase reporters (7254 runt-mut luc, 7254 ETS-mut luc, 7254 mut +4 luc) showed almost no response. pCX2neoNH was transfected into MC3T3/E1 cells. Permanent transfectants expressing dominant negative PEBP2aA were selected by G418 for 4 weeks. Two colonies were selected, and the levels of opn mRNA expression were compared between untreated MC3T3/E1 cells and G418-resistant colonies transfected with pCX2neo only which did not contain dominant negative PEBP2aA. A marked decrease in

Figure 6 Northern blotting analysis of expression of the endogenous opn gene in MC3T3/E1 cells. Dominant negative PEBP2aA plasmid, pCX2neoNH, was transfected into MC3T3/ E1 cells, and two colonies were selected (indicated as NH Neo 1, 2). A backbone plasmid harboring no cDNA sequence was also transfected into MC3T3/E1 cells (indicated as Neo). RNA was extracted from these cells and untreated MC3T3/E1 cells, and aliquots of 25 mg of total RNA were loaded per lane, blotted and hybridized with opn cDNA probe

level of endogenous opn mRNA was observed in the dominant negative PEBP2aA transfectants (Figure 6). Discussion The assembly of proteins may have important implications for the accuracy and diversity of transcriptional regulation in vivo. Thus, multiple protein-protein interactions may be indispensable to allow for the recruitment of a particular protein into a functional enhancer complex. In the promoter region of the opn gene, the runt, ets binding sites and their intervening sequences are conserved. So, these sites and their relationship may be important for transactivation of the opn gene. We investigated the eects of transcription factors, PEBP2aA and ETS1, on opn gene expression in vivo and in vitro. A putative ets binding site is located 9 nt downstream from the runt site in the promoter region of the opn gene. Among the ets family, ETS1 mRNA was known to be localized in the bone tissue (Kola et al., 1993). The signi®cance of the interaction between runt family and ets family has been reported (Giese et al., 1995; Wotton et al., 1994). PEBP2aA, ETS1 and DNA were reported to form a ternary complex and have a synergistic eect on transactivation of the T cell receptor b gene, in which runt and ets binding sites are juxtaposed. In contrast, in the opn gene runt and ets binding sites are separated by a sequence of 9 nt which corresponds to one helical turn on the DNA. An interval of 9 nt might be more favorable for stability and interaction of transcription factors which resulted synergistic transactivation than juxtaposition. It was shown that formation of a stable ternary PEBP2aADNA(oligo 3)-ETS1 complex requires the presence of


both binding sites for these proteins (Figure 5a). A ternary PEBP2aA-DNA-ETS1 complex was observed when oligo 3 +1 or oligo 3 +4 was used as a probe. However, these results could not demonstrate the direct binding between PEBP2aA and ETS1. The presence of the complex was demonstrated by using two oligonucleotides containing either the runt or ets binding site. The similar mobility of the band obtained by EMSA indicated the presence of DNA(oligo 1)-PEBP2aAETS1-DNA(oligo 2) complex. These results indicated the anity of the protein-protein interaction between PEBP2aA and ETS1 might be strengthened after binding of transcription factors to DNA. A moderate transactivation eect of the opn promoter was observed when PEBP2aA, PEBP2b or ETS1 was transfected. Previous reports showed no or little transactivation eect by PEBP2b and ETS1 alone (Ogawa et al., 1993b; Thomas et al., 1997). We considered the transactivation by PEBP2b or ETS1 occurred by cooperation with endogenously expressed PEBP2aA in the recipient cells (Figure 4). The signi®cance of PEBP2aA-PEBP2b-ETS1-DNA complex on the opn promoter activity has been suggested by the synergistic transactivation of three proteins. Opn promoter activity was markedly decreased in 7254 runt-mut luc or 7254 ets-mut luc (7254 luc containing a mutated runt or ets site, respectively). It was shown that opn promoter transactivation by PEBP2aA requires both intact runt and ets sites, and transactivation by ETS1 likewise requires both intact sites. Moreover, markedly decreased opn promoter activity was observed in 7254 mut +4 luc, in which 4 nt were inserted between runt and ets binding sites compared to 7254 luc. This 4 nt insertion changed the distance between runt and ets binding sites and the helical phasing relationship was changed by half a helical turn. Due to the loss of speci®c interactions by changing the helical phasing relationship, the synergistic transactivation system might be disrupted. The dependence on a speci®c helical phase of these binding sites also strongly suggests a functional interaction between runt and ets families. On the other hand, the oligonucleotide containing 4 nt insert between runt and ets binding sites (oligo 3 +4) showed slower mobility, similar to those of oligo 3 and oligo 3 +1 by EMSA. This result agreed with a previous report which showed that DNA binding of PEBP2aA and ETS1 was independent of the spacing between runt and ets binding sites (Wotton et al., 1994). The discrepancy between the results obtained by EMSA and transfection assay suggested that protein-DNA binding takes place due to an independent interaction between PEBP2aA-runt binding site and ETS1-ets binding site. The helical phasing relationship between the runt and ets sites in the opn promoter was found to be important for transactivation function in transfection assays, but not essential for the assembly of a protein-DNA complex in vitro. It thus appears that cooperative direct binding rather than independent binding of PEBP2aA and ETS1 is the main determinant of opn promoter function. Furthermore, direct regulation of opn gene expression by PEBP2aA has been investigated by the transfection of dominant negative PEBP2aA, pCX2neoNH, was transfected into the osteogenic cell line MC3T3/E1 which abundantly expressed opn mRNA. Marked decreased endogenous opn expression

was observed in the transfectants. The result indicate that endogenous opn is also regulated by PEBP2aA in MC3T3/E1 cells which express endogenous PEBP2aA, but not PEBP2aB or PEBP2aC as demonstrated by Northern blotting (data not shown). In homozygous mice (7/7), the calci®cation process was almost completely blocked in the skeletal tissues. However, signi®cant calci®cation was observed in the otolith in the ear (data not shown). We previously reported that the sensory hair cells of the cochlea expressed opn mRNA and that Opn protein is one of the components of the otolith. In situ hybridization demonstrated that opn mRNA is expressed in the sensory hair cells of 7/7 mice, and the level of the signal was comparable to that in normal mice (Sato et al. in preparation). Our observation strongly suggested that expression of opn mRNA in the bone is critically controlled by PEBP2aA but not in the epithelial cells of the cochlea. Opn mRNA was also detected in the distal tubules of the developing kidney and a population of macrophages in homozygous mutant mice (7/7) (data not shown). These results indicate that PEBP2aA is not essential for expression of the opn gene in tissues other than bone. We are currently investigating the expression of other members of the runt and ets families in the inner ear, epithelial cells of kidney and macrophages. We also found that opn mRNA expression in macrophages/osteoclasts was markedly decreased in mice carrying double alleles of mi mutants (Sato et al. in preparation). The bHLH transcription factor (MITF) is encoded in the mi locus. Interactions between MITF and members of runt and ets families should be investigated further. Opn protein is considered to be closely related to calci®cation. Our previous in vivo studies also showed that opn mRNA expression is closely related to physiological and pathological calci®cation (Hirota et al., 1993, 1995; Nakase et al., 1994). We considered that the defective calci®cation in the homozygous (7/7) mice was due to the lack of opn gene expression. Previous reports indicated that PEBP2aA regulated not only the opn gene but also BSP, osteocalcin and type I collagen gene in vitro (Ducy et al., 1997). In the present study, direct regulation of opn gene expression controlled by PEBP2aA and ETS1 was shown. We consider the defective osteogenesis observed in the homozygous (7/7) mice to be a result of decreased opn gene expression. In normal mice, abundant opn mRNA but not BSP mRNA expression was demonstrated in the hypertrophic chondrocytes located in the provisional calci®cation zone. Furthermore, the bone tissues of knockout mice lacking BSP gene expression were normal (Aubin et al., 1996). Increased bone formation was demonstrated in the osteocalcin knockout mice (Ducy et al., 1996). Therefore, BSP and osteocalcin are unlikely to be the target genes in vivo which resulted in the phenotype of homozygous (7/7) mice. The phenotype of osteogenesis imperfecta due to mutated type I collagen gene is dierent from that seen in cleidocranial dysplasia due to lack of PEBP2aA gene products. Furthermore, type I collagen mRNA was detected in 7/7 mice (data not shown). Therefore type I collagen

1523


1524

may not be the target gene in vivo which resulted in the phenotype of homozygous (7/7) mice. We are currently generating knockout mice completely lacking of opn mRNA and transgenic mice carrying a dominant negative opn gene. Comparison of the phenotypes of these and PEBP2aA knockout mice should allow further elucidation of the mechanism of calci®cation in vivo. Materials and methods Cell culture The NIH3T3 ®broblast cell line was provided by Dr SA Aaronson (National Cancer Institute, Bethesda, MD) and maintained in Dulbecco's modi®ed Eagle's medium (DMEM; ICN Biomedicals) supplemented with 10% fetal calf serum (FCS; Nippon Bio-supp Center, Tokyo, Japan). The MC3T3/E1 osteoblast cell line was generously provided by Dr Kodama (Sudo et al., 1983) and maintained in alpha modi®ed of Eagle's minimal essential medium (a-MEM; ICN Biomedicals, Costa Mesa, CA) supplemented with 10% FCS. The 293T human embryonic kidney cell line was kindly provided by Dr D Baltimore (Rockefeller University, New York, NY) and maintained in DMEM supplemented with 10% FCS. Northern blotting Total RNAs were prepared by the lithium chloride-urea method (Auray and Rougeon, 1980). The mouse opn cDNA containing a 1.2 kb fragment (Nomura et al., 1988), PEBP2aA and ETS1 (Yamada et al., 1995) labeled with 32 P-a-dCTP (DuPont/NEN Research Products, Boston, MA; 10 mc/ml) by random oligonucleotide priming were used as probes. After hybridization at 488C, blots were washed to a ®nal stringency of 0.26SSC (16SSC is 150 mM NaCl, 15 mM trisodium citrate, pH 7.4) at 508C and subjected to autoradiography. In situ hybridization Wild-type (+/+) mice embryos at 18.5 days post coitum (pc) were ®xed with 4% paraformaldehyde in 100 mM PB (pH 7.4) overnight, and embedded in paran. The preparation of opn and PEBP2aA, B, C probes and in situ hybridization were performed as described previously (Satake et al., 1995). Construction of eector and reporter plasmids pEF-BOS expression vector was kindly provided by Dr S Nagata (Mizushima and Nagata, 1990). The Bg1II ± MluI fragment from PEBP2aA cDNA was cloned into the blunted XbaI site of the pEF-BOS. The luciferase gene subcloned into pSP72 (pSPLuc) was generously provided by Dr K Nakajima (Nakajima et al., 1993). To construct reporter plasmids, a DNA fragment containing a promoter region and the part of the ®rst exon of the opn gene (nt 7910 to +66) was cloned upstream of the luciferase gene into pSPLuc. The deletion constructs of the opn promoter region generated by PCR were also ligated into pSPLuc. pCX2neoNH, dominant negative PEBP2aA, contains the NCoI ± HindIII fragment (nt 1323 to 1718; the NH fragment) of PEBP2aA cDNA that spans the Runt domain of PEBP2aA. (Sakakura et al., 1994).

Transient co-transfection assay NIH3T3 cells (56105) were plated in 10 cm dishes 1 day before transfection. Ten mg of reporter, 10 mg of eector and 1 mg of expression vector containing the b-galactosidase gene (Tokunaga et al., 1986), which was used to evaluate transfection eciency, were mixed and cotransfected by the calcium phosphate precipitation method (Graham and Van der Eb, 1973). The cells were collected 48 h after transfection and lysed with 100 mM potassium phosphate buer (pH 7.4) containing 1% Triton X-100 (Sigma). Soluble extracts were then assayed for luciferase activity with a luminometer LB96P (Berthold GmbH, Wildbad, Germany) and for bgalactosidase activity. The luciferase activity was normalized by the galactosidase activity and total protein concentration according to the method described by Yasumoto et al. (1994). The normalized value was divided by that obtained with co-transfection of the reporter and pEF-BOS, and is expressed as the relative luciferase activity. Introduction of dominant negative PEBP2aA into an osteogenic cell line MC3T3/E1 cells (56105) were plated in 10 cm dishes 1 day before transfection. Transfection of 20 mg of pCX2neoNH was performed by the calcium phosphate precipitation method (Graham and Van der Eb, 1973). After 3 days of culture, fresh medium containing 600 mg of G418 per ml was added. G418-resistant cells were isolated after 4 weeks of selection. Electrophoretic mobility shift assay (EMSA) 293T cells were transfected with a PEBP2aA expression plasmid. Whole-cell extracts were prepared 48 h after DNA transfection by a freezing-thawing method as previously described (Pagano et al., 1992) and subjected to EMSA. In vitro translated ETS1 was prepared using an ETS1 cDNA in the pSG5 vector (Stratagene). In vitro transcription and translation (using rabbit reticulocytes) was performed according to the manufacturer's instructions (Promega, Madison, WI) (Nimer et al., 1996). Oligonucleotides were labeled with 32P-a-dCTP by ®lling in 5'-overhangs and used as probes for EMSA. DNA-binding assays were performed in a 20 ml reaction mixture containing 10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 75 mM KCl, 1 mM dithiothreitol, 4% Ficoll type 400, 500 ng of poly (dl-dC), 25 ng of labeled DNA probe and 1 mg whole-cell extract of 1 ml in vitro transcribed and translated recombinant protein. After incubation at room temperature for 30 min, the reaction mixture was subjected to electrophoresis at 14 V/cm at 48C on 5% polyacrylamide gels in 0.256TBE buer (16TBE is 90 mM Tris-HCl, 64.6 mM boric acid and 2.5 mM EDTA, pH 8.3). The polyacrylamide gels were dried on Whatman 3MM chromatography paper and subjected to autoradiography.

Acknowledgements We thank A Fukuyama and K Morihana for their excellent technical support. This work was supported by grants from the Ministry of Education, Science and Culture and the Ministry of Health and Welfare.


1525

References Aubin JE, Gupta AK, Zirngbl R and Rossant J. (1996). J. Bone. Miner. Res., 11, S102. Auray C and Rougeon F. (1980). Eur. J. Biochem., 107, 303 ± 314. Bae SC, Yamaguchi-Iwai Y, Ogawa E, Maruyama M, Inuzuka M, Kagoshima H, Shigesada K, Satake M and Ito Y. (1993). Oncogene, 8, 809 ± 814. Bae SC, Takahashi E, Zhang YW, Ogawa E, Shigesada K, Namba Y, Satake M and Ito Y. (1995). Gene, 159, 245 ± 248. Brown LF, Papadopoulos SA, Berse B, Manseau EJ, Tognazzi K, Perruzzi CA, Dvorak HF and Senger DR. (1994). Am. J. Pathol., 145, 610 ± 623. Cameron S, Taylo DS, TaPas EC, Speck NA and MathetPrevot B. (1994). Blood, 83, 2851 ± 2859. Craig AM and Denhardt DT. (1991). Gene, 100, 163 ± 171. Denhardt DT and Guo X. (1993). FASEB J., 7, 1475 ± 1482. Ducy P, Desbois C, Boyce B, Pinero G, Story B, Dunstan C, Smith E, Bonadio J, Goldstein S, Gundberg C, Bradley A and Karsenty G. (1996). Nature, 382, 448 ± 452. Ducy P, Zhang R, Georoy V, Ridall AL and Karsenty G. (1997). Cell, 89, 747 ± 754. Frank R, Zhang J, Uchida H, Meyers S, Hiebert SW and Nimer SD. (1995). Oncogene, 11, 2667 ± 2674. Giese K, Kingsley C, Kirshner JR and Grosschedl R. (1995). Genes. Dev., 9, 995 ± 1008. Graham FL and Van der Eb AJ. (1973). Virology, 52, 456 ± 467. Guo X, Zhang YP, Mitchell DA, Denhardt DT and Chambers AF. (1995). Mol. Cell. Biol., 15, 476 ± 487. Hallberg B, Thornell A, Holm M and Grundstrom T. (1992). Nucleic Acids Res., 20, 6495 ± 6499. Hirakawa K, Hiorta S, Ikeda T, Yamaguchi A, Takemura T, Nagoshi J, Yoshiki S, Suda T, Kitamura Y and Nomura S. (1994). J. Bone. Miner. Res., 9, 1551 ± 1557. Hirota S, Imakita M, Kohri K, Ito A, Morii E, Kim H-M, Adachi S, Kitamura Y and Nomura S. (1993). Am. J. Pathol., 143, 1003 ± 1008. Hirota S, Ito A, Nagoshi J, Takeda M, Kurata A, Takatsuka Y, Kohri K, Nomura S and Kitamura Y. (1995). Lab. Invest., 72, 64 ± 69. Hsiang YH, Spencer D, Wan S, Speck NA and Raulet DH. (1993). J. Immunol., 150, 3905 ± 3916. Kagoshima H, Satake M, Miyoshi H, Ohki M, Pepling M, Gergen P, Shigesada K and Ito Y. (1993). Trends Genetics, 9, 338 ± 341. Kamachi Y, Ogawa E, Asano M, Ishida S, Murakaml Y, Satake M, Ito Y and Shigesada K. (1990). J. Virol., 64, 4808 ± 4819. Kohri K, Nomura S, Kitamura Y, Nagata T, Yoshioka K, Iguchi M, Yamate T, Umekawa T, Suzuki Y, Shinohara H and Kurita T. (1993). J. Biol. Chem., 268, 15180 ± 15184. Kola I, Brookes S, Green AR, Garber R, Tymms M, Papas TS and Seth A. (1993). Proc. Natl. Acad. Sci. USA, 90, 7588 ± 7592. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, Deguchi K, Shimizu Y, Bronson RT, Gao Y-H, Inada M, Sato M, Okamoto R, Kitamura Y, Yoshiki S and Kishimoto T. (1997). Cell, 89, 755 ± 764. Levanon D, Negreanu V, Bernstein Y, Baram I, Avivi L and Groner Y. (1994). Genomics, 23, 425 ± 432. Mizushima S and Nagata S. (1990). Nucl. Acids. Res., 18, 5322. Nakajima K, Kusafuka T, Takeda T, Fujitani Y, Nakae K and Hirano T. (1993). Mol. Cell. Biol., 13, 3027 ± 3041. Nakase T, Takaoka K, Hirakawa K, Hirota S, Takemura T, Onoue H, Takebayashi K, Kitamura Y and Nomura S. (1994). Bone Min., 26, 109 ± 122. Nimer SD, Zhang W, Kwan K, Wang Y and Zhang J. (1996). Blood, 87, 3694 ± 3703.

Nomura S, Wills AI, Edwards DR, Heath IK and Hogan BML. (1988). J. Cell. Biol., 106, 441 ± 450. Nuchprayoon I, Meyers S, Scott LM, Suzow J, Hiebert S and Friedman AD. (1994). Mol. Cell. Biol., 14, 5558 ± 5568. Oates AJ, Barraclough R and Rudland PS. (1996). Oncogene, 13, 97 ± 104. Ogawa E, Maruyama M, Kagoshlma H, Inuzuka M, Lu J, Satake M, Shigesada K and Ito Y. (1993a). Proc. Natl. Acad. Sci. USA, 90, 6859 ± 6863. Ogawa E, Inuzaka M, Manuyama M, Satake M, NaltoFujlmoto M, Ito Y and Shigesada K. (1993b). Virology, 194, 314 ± 331. Oldberg A, Franzen A and Heinegard D. (1986). Proc. Natl. Acad. Sci. USA, 83, 8819 ± 8823. Patarca R, Freeman GJ, Sing RP, Wei FY, Durfee T, Blattner F, Regnier DC, Kozak CA, Mock BA, Morese HCIII, Jerrells TR and Cantor H. (1989). J. Exp. Med., 170, 145 ± 162. Pagano M, Durst M, Joswig S, Draetta G and Jansen-Durr P. (1992). Oncogene, 7, 1681 ± 1686. Prosser HM, Wotton D, Gegonne A, Ghysdael J, Wang S, Speck NA and Owen MJ. (1992). Proc. Natl. Acad. Sci. USA, 89, 9934 ± 9938. Redondo JM, Pfohl JL, Hernandez-Munain C, Wang S, Speck NA and Krangel M. (1992). Mol. Cell. Biol., 12, 4817 ± 4823. Sakakura C, Yamaguchi-Iwai Y, Satake M, Bae SC, Takahashi A, Ogawa E, Hagiwara A, Takahashi T, Murakami A, Makino K, Nakagawa T, Kamada N and Ito Y. (1994). Proc. Natl. Acad. Sci. USA, 91, 11723 ± 11727. Satake M, Nomura S, Yamaguchi-Iwai Y, Takahama Y, Hashimoto Y, Niki M, Kitamura Y and Ito Y. (1995). Mol. Cell. Biol., 15, 1662 ± 1670. Sato M, Yasui N, Nakase T, Kawahata H, Sugimoto M, Hirota S, Kitamura Y, Nomura S and Ochi T. J. Bone. Miner. Res., in press. Singh R, Patarca R, Schwartz J, Singh P and Cantor H. (1991). J. Exp. Med., 171, 1931 ± 1942. Sudo H, Kodama H, Amagai Y, Yamamoto S and Kasai S. (1983). J. Cell. Biol., 96, 191 ± 198. Takahashi A, Satake M, Yamaguchi-Iwai Y, Bae SC, Lu J, Maruyama M, Zhang YW, Oka H, Arai N, Arai K and Ito Y. (1995). Blood, 86, 607 ± 616. Takemura T, Sakagami M, Nakase T, Kubo T, Kitamura Y and Nomura S. (1994). Hearing Res., 79, 99 ± 104. Thomas RS, Tymms MJ, McKinlay LH, Shannon MF, Seth A and Kola I. (1997). Oncogene, 14, 2845 ± 2855. Tokunaga K, Taniguchi H, Yoda K, Shimizu M and Sakiyama S. (1986). Nucl. Acids Res., 14, 2829. Wang S and Speck NA. (1992). Mol. Cell. Biol., 12, 89 ± 102. Wang S, Wang O, Crute BE, Melnikova IN, Keller SR and Speck NA. (1993). Mol. Cell. Biol., 13, 3324 ± 3339. Wargnier A, Legnos-Maida S, Bosselut R, Bourge JF, Lafaurie C, Ghysdael CJ, Sasportes M and Paul P. (1995). Proc. Natl. Acad. Sci. USA, 92, 6930 ± 6934. Weber GF, Ashkar S, Glimcher MJ and Cantor H. (1996). Science, 271, 509 ± 512. Wijmenga C, Speck NA, Dracopoli NC, Hofker MH, Liu P and Collins FS. (1995). Genomics, 26, 611 ± 614. Wotton D, Ghysdael J, Wang S, Speck NA and Owen MJ. (1994). Mol. Cell. Biol., 14, 840 ± 850. Yamada T, Hitomi Y, Satake M and Oikawa T. (1995). Eur. J. Immunol., 25, 2710 ± 2713. Yasumoto K, Yokoyama K, Shibata K, Tomita Y and Shibahara S. (1994). Mol. Cell. Biol., 14, 8058 ± 8070. Zhang DE, Fujioka K, Hetherington CJ, Shapiro LH, Chen HM, Look AT and Tenen DG. (1994). Mol. Cell. Biol., 14, 8085 ± 8095.